皮盼资讯网移动版

本文使用项目K主题评级矩阵来计算相似性，针对某些内容，解决由稀疏用户项评级矩阵引起的数据稀疏性问题，并将计算维度从用户数减少到主题数。项目K主题评级矩阵描述了来自K方面的餐厅的特征，其比用户评级矩阵更详细，因此带来更准确的相似性和评级预测。

6. 结论

本文从挖掘用户关心的多种潜在主题的角度研究了餐厅推荐系统。我们设计了潜在主题分布特征和项目-K主题评级矩阵，以整合潜在因素和评级记录中包含的特征，并将它们应用于相似度计算。本文还进行了一项实验，以探索最佳主题数K，并将性能与基线算法-Item CF进行比较。根据实验结果，我们可以得出结论，挖掘用户关心的潜在方面可以帮助解决餐厅推荐的问题。

7. 参考文献

[1] Lihua Sun, Junpeng Guo, Yanlin Zhu. Applying uncertainty theory into the restaurant recommender system based on sentiment analysis of online Chinese reviews[J]. World Wide Web,2019,2019, 22(1): 83-100

[2] Sonya Zhang, Mohammad Salehan, Andrew Leung, Ishmene Cabral, Navid Aghakhani. A Recommender System for Cultural Restaurants Based on Review Factors and Review Sentiment[A]. AMCIS[C].2018

[3] Chao Li, Sen Feng, Qingtian Zeng, Weijian Ni, Hua Zhao, Hua Duan:Mining Dynamics of Research Topics Based on the Combined LDA and WordNet[J]. IEEE Access,2019,7: 6386-6399

[4] Maha Amami, Gabriella Pasi, Fabio Stella, Rim Faiz.An LDA-Based Approach to Scientific Paper Recommendation[J]. NLDB,2019: 200-210

[5] Shinjee Pyo, Eunhui Kim, Munchurl Kim. LDA-Based Unified Topic Modeling for Similar TV User Grouping and TV Program Recommendation[J]. IEEE Trans. Cybernetics,2015,45(8): 1476-1490

[6] Yifan Gao, Wenzhe Yu, Pingfu Chao, Rong Zhang, Aoying Zhou, Xiaoyan Yang:A Restaurant Recommendation System by Analyzing Ratings and Aspects in Reviews. DASFAA,2015,(2) : 526-530

[7] Koren Y , Sill J. Proceedings of the Twenty-Third international joint conference on Artificial Intelligence[A].Collaborative filtering on ordinal user feedback[C]. 2013. 

(责编：刘扬、赵光霞)